Figure 7
Accuracy of movement can be achieved through calibration. For this typical calibration data set, the fitted polynomial functions all have means for the error residuals that are not significantly different from zero. However, the residuals for lower-order (2–5) and middle-order (6–8) polynomials usually fail the assumption of normality. Higher-order polynomials show no trend over the full length of the stage-position encoder. As the numbers of orders included increases (8–12), the variation of the fitted model to the real-world coordinates becomes acceptable for the current application (4th, SD 0.4 mm; 8th, SD 0.2 mm, 12th, SD 0.1 mm). |